Fast Atom Bombardment Mass Spectrometry and Tandem Mass Spectrometry of Biologically Active Peptidoglycan Monomers from Neisseria gonorrhoeae” mass spectrometry

atom bombardment monomers, providing molecular weight and partial structural information. Approximately 0.5-0.7 p1 of the glycerol matrix containing the sample (see above) was applied to a stainless steel sample stage mounted on the end of a high vacuum push rod. The sample was inserted via vacuum locks into the center of the ion source where a neutral xenon beam (10 FA, 7 kV, Ion Tech B12N neutral source, Teddington, UK) impinged upon the matrix containing the sample. Ions produced by the interaction of the neutral beam with the sample surface were accelerated, energy and mass selected, and detected with a secondary electron multiplier. Exact mass measurements were made in the peak matching mode (at a resolution of 1/10,000) employing I1 (Beckman Biochemical) protonated molec- ular ion (MH+) 926.5212 coupound Collisions of the precursor ion with the neutral gas convert a fraction of the translational energy of the precursor into vibrational energy resulting in bond cleavage. The fragment ions produced from these bond cleavages are referred to as product ions. Scanning MS-2 results in a mass spectrum (product ion spectrum) which contains only information relating to the compound whose mass was selected in MS-1 and can, therefore, be used to deduce its structure. In the experiments discussed below, the resolution of MS-1 and MS-2 was 1/1000 which resulted in unit resolution of both the precursor and product ion mass spectra (50). Reported masses for both the structures and mass spectra are rounded down to the nearest integral mass for clarity. The data system-assigned mass values differed by <+0.3 daltons from that calculated (to one decimal point) for the proposed fragment components.

(1). Peptide cross-linking bonds between amino acid residues located on different glycan chains lead to the formation of a complex three-dimensional macromolecule that has been likened to an enormous, covalently closed basket surrounding the cytoplasmic membrane (2). Although nature has provided numerous subtle variations in the composition of peptidoglycan among the bacteria (3), this rather rigid arrangement of polymeric glycan (up to 100 disaccharide units long) crosslinked by peptides has been remarkably well conserved, a fact undoubtedly related to its role in maintaining the physical integrity of the bacterial cell. Yet, when taken from the host's perspective, peptidoglycan is more than merely a biologically inert bacterial corset. Indeed, given access to host tissues and cells, soluble peptidoglycan derivatives are proving to be versatile biological effectors which, as a class, have a propensity to modulate immune and inflammatory reactions. Among the numerous peptidoglycan-mediated activities that have been well documented in recent years are adjuvanticity (4, 5 ) , pyrogenicity (6,7), activation of the metabolic and killing capacity of macrophages (8,91, stimulation of leukocytes to release pharmacologically active mediators including interleukin-1 (10, ll), and arthritogenicity (12,13). Very recently, certain peptidoglycan fragments have even been implicated as naturally occurring neuromodulators based on data showing that they accumulate in the brains and urine of sleepdeprived animals and induce excess slow-wave sleep (14,15).
During the past several years, we have been testing the hypothesis that peptidoglycan fragments influence the host response during the natural course of bacterial infections. Toward this end, we have exploited Neisseria gonorrhoeae as a model organism in which peptidoglycan-host interactions might be particularly direct and extensive in vivo (16)(17)(18)(19)(20). To date, we have identified several sets of purified gonococcal peptidoglycan fragments that likely gain access to host tissues; these range from high molecular weight (>lo6 daltons) soluble fragments that are extensively substituted in the glycan with 0-acetyl derivatives (19, 21, 22) to unusual anhydromuramic acid-containing disaccharide peptide monomers (-lo3 daltons), the major peptidoglycan compounds released by growing gonococci (16)(17)(18). Collectively, these gonococcal peptidoglycan fragments have been found to mediate diverse biological activities including arthritogenicity (23), toxicity for human fallopian tube mucosa (24), and complement activation (25). However, our aim to determine the structural requirements and molecular mechanisms of these activities is potentially compromised by the outdated procedures previously employed for purification and chemical analysis of peptidoglycan fragments. In fact, the whole field of peptidoglycan chemistry has, until recently, changed very little since the classical approaches offered by Ghuysen (26) and others dur-

Mass Spectrometry
of Peptidoglycan Monomers 7515 ing the late 1960s. There are several reasons for these difficulties in peptidoglycan chemistry that are related primarily to the unique structure of peptidoglycan, e.g. the relative inapplicability of peptide sequencing techniques to the fine structure analysis of peptidoglycan and the inability to hydrolyze selectively the various positions of the peptide side chain of peptidoglycan. Even the exemplary work of van Heijenoort and co-workers (27), applying conventional mass spectrometry to the analysis of derivatized peptidoglycan disaccharides containing 1,6-anhydro-N-acetylmuramic acid, has not been generally applicable to the study of the disaccharide peptide fragments of peptidoglycan.
Fortunately, vastly improved methodology for peptidoglycan chemistry is now being developed. First is the successful fractionation of low molecular weight peptidoglycan fragments by reverse phase high performance liquid chromatography (HPLC) as evinced in the original report of Glauner and Schwarz (28) and in the applications of Daugherty (29) and Martin et al. (30). These studies have demonstrated that muramidase digests of peptidoglycan from at least certain bacteria are considerably more complex chemically than previously appreciated. Yet, even with these excellent means of separation, the analysis by classical techniques of the numerous peptidoglycan products resolved requires a rather heroic effort and is indirect, thus typically falling well short of anything resembling unambiguous proof of structure.
The introduction of fast atom bombardment mass spectrometry (FABMS) to the early 1980s (31,32), its application to peptidoglycans (30,33,34), and the more recent commercial availability of tandem magnetic four-sector mass spectrometers (35) seem to offer a novel approach to this problem by providing an efficient and unambiguous determination of the molecular weight and primary structure of peptidoglycan fragments. For conventional peptides FABMS produces primarily molecular weight information. Depending on several factors, including sample concentration and composition, some fragment ions indicative of the primary structure of the peptide may be observed and provide sequence information (36, 37), but their relative abundance is typically 5-10-fold less than that of the molecular ion. There are, however, several factors, including matrix interferences and the presence of other components in a mixture, which may severely limit the extent of sequence information derived from the sample (38). These observations also hold true for other compound classes, including peptidoglycan. The application of tandem mass spectrometry (MS/MS) to the analysis of peptidoglycan should overcome these limitations and increase the structural information available. In MS/MS the ions associated with the molecular weight of the compound of interest are selected in the first mass spectrometer (MS-1) at a resolution of one mass unit. These ions, which uniquely define the sample, collide with an inert gas such as helium, producing fragment ions which are mass analyzed in the second mass spectrometer, MS-2. The result of this two-stage process is a mass spectrum rich in structural information related only to the compound selected in MS-1 (39).
Accordingly, to study the structure-function relationship of activities mediated by gonococcal peptidoglycan fragments, we have used reversed phase HPLC, FABMS, and MS/MS to define the numerous analogs comprising two distinct enzymatically derived classes of disaccharide peptide monomers isolated from gonococcal peptidoglycan. Each of these classes, Le. "Chalaropsis monomers" with reducing muramic acid ends and "anhydro monomers" with nonreducing 1,6-anhydromuramic acid ends, is biologically active in one or more experi-mental systems of interest (23,24), and each likely interacts with host tissues in vivo (18,19,40).
Preparation of Peptidoglycan Monomers-Purified intact peptidoglycan was used as starting material for two structurally related families of monomeric peptidoglycan fragments. Each of these sets, which are referred to as Chalaropsis monomers and anhydro monomers, respectively, was initially isolated as mixtures of peptidoglycan monomers. Chalaropsis monomers were isolated by gel filtration on connected columns of Sephadex (3-50 and G-25 after complete digestion with Chalaropsis B muramidase (Miles Laboratories, Elkhart, IN) of intact extensively 0-acetylated peptidoglycan from strain FA19 or of 0-acetyl-deficient peptidoglycan from strain RD5, as we have described previously (24,40). Pooled monomeric fractions were desalted by gel filtration on Sephadex G-15 eluted with pyrogen-free water. Chalaropsis monomers served as the source of peptidoglycan monomers with hydrated, reducing N-acetylmuramic acid ends.
Anhydro monomers were prepared from intact strain RD5 peptidoglycan with use of a partially purified enzyme preparation obtained from Escherichia coli ATCC 9637. This preparation contained both DD-endopeptidase and peptidoglycan:peptidoglycan-6-muramyl transferase (transglycosylase) activities. Several closely related procedures for purification of these peptidoglycan hydrolases have been reported (44)(45)(46). For our purposes, the optimal procedure was a variation of this basic method' in which the key step involved chromatography on carboxymethyl-Sepharose CL-GB (Pharmacia P-L Biochemicals) of Triton X-100 extracts of sonicated E. coli. This chromatographic procedure was performed as described previously (22) except that Triton X-100 extracts of washed membranes (rather than extracts of combined cytoplasmic plus membrane fractions) served as the source of the enzymatic activity. Details of the protocol for the complete digestion of intact peptidoglycan with the E. coli transglycosylase-endopeptidase have been published (24). Anhydro monomers were isolated from the peptidoglycan digest by gel filtration and desalted as for Chalaropsis monomers. Using this procedure, the yield of anhydro monomers from intact peptidoglycan starting material was exceptionally high (-60%). The efficiency of this reaction was attributed to the virtually complete conversion of insoluble peptidoglycan to anhydro monomers by the novel use of an enzyme preparation which contained both glycan-splitting (transglycosylase) and peptide-splitting (endopeptidase) activities. Anhydro monomers served as the source of peptidoglycan monomers with nonreducing 1,6-anhydro-N-acetylmuramic acid ends.
Previous studies (18,24,40) employing traditional procedures for peptidoglycan chemistry have revealed that the major components of Chalaropsis monomers were N-acetylglucosaminyl-N-acetylmuramylalanyl-glutamyl diaminopimelic acid and the corresponding disaccharide tetrapeptide with a COOH-terminal alanine. Anhydro monomers were composed predominantly of the respective disaccharide peptides containing the 1,6-anhydromuramic acid end.
High Performance Liquid Chromatography-Final purification of the individual components of Chalaropsis monomers and anhydro monomers was accomplished by reversed phase HPLC. Samples were separated using a Waters HPLC 510 binary pumping system with solvent programmer and reversed phase columns. A Waters 491 _____ " " * U. Schwarz, personal communication.
absorbance detector operated at 214 nm and a Hewlett-Packard 3390A integrator were used for detection. Initial separation employed 4 X 250-mm columns from Vydac (C, and CIS, The Separations Group, Hesperia, CA) and Waters (Cis, Millipore Corp., Milford, MA). Several HPLC fractions were subjected to further chromatography using a Waters narrowbore (2 X 250-mm) reversed phase C1, column to improve chromatographic resolution. The solvent gradient employed depended on the column, the complexity of the sample, and the resolution of the separation. Typical elution gradients were linear from 100% water (containing 0.050% CF3COOH) to 25% CH&N (containing 0.035% CF&OOH) over a period of 30 min. The flow rates were 1.0-1.5 ml/min for the 4 X 250-mm columns and 0.4 ml/ min for the narrowbore column.
Sample Preparation-The samples for FAB must be dissolved in a liquid matrix (38, 47,48) in order to observe abundant long-lasting secondary ion signals associated with the species of interest. Glycerol, which is the most widely used matrix for the analysis of biological molecules, was employed in the analysis of peptidoglycan monomers. The peptidoglycan monomer mixtures or individual fractions isolated by HPLC were placed in 1-ml conical vials to which glycerol and 30% aqueous acetic acid were added in volume ratio of 5:l. Sample concentrations ranged from 0.1 to 10.0 nmol/rl of matrix with a total matrix volume of 3-5 pl.
Fast Atom Bombardment Mass Spectrometry-A double focusing (Finnigan MAT 731, Bremen, FRG) mass spectrometer of the Mattauch-Herzog geometry (38) and a tandem mass spectrometer (JEOL HX110/HX110, Tokyo) were employed in this work (35). The MAT 731 has a mass range of 2000 daltons at 8-kV accelerating potential and was employed for the initial characterization of peptidoglycan monomers, providing molecular weight and partial structural information. Approximately 0.5-0.7 p1 of the glycerol matrix containing the sample (see above) was applied to a stainless steel sample stage mounted on the end of a high vacuum push rod. The sample was inserted via vacuum locks into the center of the ion source where a neutral xenon beam (10 FA, 7 kV, Ion Tech B12N neutral source, Teddington, UK) impinged upon the matrix containing the sample. Ions produced by the interaction of the neutral beam with the sample surface were accelerated, energy and mass selected, and detected with a secondary electron multiplier. Exact mass measurements were made in the peak matching mode (at a resolution of 1/10,000) employing [Sar'-Alaa]angiotensin I1 (Beckman Biochemical) protonated molecular ion (MH+) 926.5212 as a reference coupound mixed with the sample on the probe tip at a ratio (w/w) of 1:3.
The JEOL HX110/HX110 tandem mass spectrometer consists of two consecutive double focusing mass spectrometers (MS-1 and MS-2) each employing an electric field (E) followed by a magnetic field (B) (49), i.e. an EBEB geometry and a mass range of 14,500 daltons at 10-kV accelerating potential. The methods of sample preparation, introduction, ionization, and detection were similar to those described for the MAT 731. The only difference is the use of a JEOL ion/ neutral beam source (10 mA, 6 kV) to produce the xenon primary beam which strikes the sample probe. As in the case of the MAT 731 (double focusing mass spectrometer), the JEOL HX110/HX110 may be operated using only the first EB segment (MS-1) for recording the molecular weight and full mass spectra of the samples. The unique aspect of this instrument is that it may also be operated in the MS/ MS mode. In this mode of operation, MS-1 is set to transmit only the ion of interest, generally the protonated molecular ion of the compound under investigation. These ions enter a region between MS-1 and MS-2 containing helium gas. Collisions of the precursor ion with the neutral gas convert a fraction of the translational energy of the precursor into vibrational energy resulting in bond cleavage. The fragment ions produced from these bond cleavages are referred to as product ions. Scanning MS-2 results in a mass spectrum (product ion spectrum) which contains only information relating to the compound whose mass was selected in MS-1 and can, therefore, be used to deduce its structure. In the experiments discussed below, the resolution of MS-1 and MS-2 was 1/1000 which resulted in unit resolution of both the precursor and product ion mass spectra (50).
Reported masses for both the structures and mass spectra are rounded down to the nearest integral mass for clarity. The data system-assigned mass values differed by <+0.3 daltons from that calculated (to one decimal point) for the proposed fragment components.

RESULTS
Fast Atom Bombardment Mass Spectrometry-characterization of Chalaropsis monomers and anhydro monomers of N. gonorrhoeae strain RD5 by FABMS prior to HPLC fractionation indicated the presence of several components in each preparation. The probability that all species in such a mixture could be detected by FABMS depends on several factors. First, as the number of components in the mixture increases, those compounds which are present at low molar concentrations compared to the major components are masked by the latter's abundant ion signal and the matrix background, which consists of ions from the sample and liquid matrix. Second, those compounds which are more surface active, i.e. have hydrophobic substituents, will be preferentially ionized (51). Third, in mixtures in which the various species differ by single amino acids or minor structural modifications such as cyclization with corresponding loss of HzO, it is frequently difficult to ascertain whether the observed ion is a protonated molecular ion or a fragment ion of a molecule of higher mass. The latter possibility is a prime consideration in the analysis of peptidoglycan monomers in which various structures differ by single amino acids or sugar residues and/ or HzO. Therefore, to determine the total number of com-
Separated by HPLC into single component fractions for structure function studies (15,54,55). Exact mass measurement, Table 11. e The structure of the disaccharide portion of this compound isolated from E. coli and Salmonella typhi was determined by electron ionization mass spectrometry previously by Taylor et al. (27) after derivatization. sitions of three peptidoglycan monomers were determined by exact mass measurements (Table 11) as a further confirmation of their structure. It should be noted that several compounds were detected unambiguously only after HPLC purification. These include the compound of MH' 719 (Table I) which is masked in the mixture by an abundant fragment ion in the FAB mass spectrum of MH' 922 ( Fig. IA) and MH' 908 and 979 (Table I) Table I because its structure has not been proven unambiguously) which was always observed at m/z 1036 would correspond to MH' 922 + asparagine. In addition, HPLC separation in conjunction with FABMS and MS/MS revealed an anhydro peptidoglycan monomer (MH' 851, Table I) as a minor component in the Chularopsis monomer preparation. Pre-separation by HPLC pounds in each mixture, Chalaropsis and anhydro peptidoglycan monomers were separated by HPLC, and each fraction was analyzed by FABMS. Table I    The numerical values refer to the mass of the fragments produced by cleavage along the bonds indicated plus 1 or 2 if the notation +H or +2H indicates hydrogens transferred to the species which is observed in the mass spectra. removed any ambiguity concerning the nature of these molecular ions. Recently, the presence of anhydromuramic acidcontaining monomers was detected in muramidase digests of gonococcal peptidoglycan (57).
The HPLC separation of Chalaropsis monomers was complicated by the fact that two peaks are observed for each species due to the a/@ interconversion at C-1 of MurNAc (30,52). The samples could not be reduced with sodium borohydride to the open form of MurNAc, which would have simplified the HPLC peak profile (28), because the separated, fully characterized compounds had to be tested for their ability to induce slow-wave sleep (53,54). The anhydro monomers gave a single peak by HPLC for each compound because the a l p interconversion is blocked by the 1,6-anhydro linkage (30).
In addition to molecular weight information, FABMS may provide some sequence information dependent on several factors including sample concentration, composition, and purity. To increase the probability of observing fragment ions related to specific peptidoglycan monomers, HPLC fractions containing single components were analyzed by FABMS a t concentrations of 10-15 nmol/pl of liquid matrix. Due to the amount of material required to generate fragment ions of sufficient abundance to verify the sequence, only the major components of Chalaropsis and anhydro peptidoglycan monomers, specifically MH' 851, 869, 922, 940, 993, and 1011 (Table I), were successfully characterized. An example of the nature and extent of fragmentation encountered in the characterization of peptidoglycan by FABMS is GlcNAc-(1,6anhydro)-MurNAc-Ala-Glu-A2pm, MH' 922 (Fig. 1A).
The peak at m/z 719 (= 717 + 2) corresponds to the cleavage of the disaccharide linkage with retention of the proton on the portion containing the peptide and simultaneous rearrangement of a hydrogen atom from the terminal GlcNAc (Fig. 2). The pair of peaks at m/z 532 and 534 and the peak at m/z 517 are due to the loss of the disaccharide portion of the molecule, leaving the peptide portion intact. The mass difference between these ions (mlz 719, 534, and 517) and MH' 922 identifies the carbohydrate moiety: the mass difference between MH' 922 and m/z 719 (203 daltons) corresponds to the loss of GlcNAc, whereas the difference between m/z 719 and m/z 534 (185 daltons) is due to the loss of the 1-6anhydro sugar. In the related Chalaropsis peptidoglycan monomer, MH' 940, both mass differences are 203 daltons, indicating loss of GlcNAc and the GlcNAc component of MurNAc. Furthermore, the ion at m/z 204 in Fig. 1 is characteristic of GlcNAc. Similarly, the sequence of the peptide portion can be deduced from the ions associated with the cleavage along the peptide backbone. The types of cleavages observed are identical to those found in peptides which do not contain carbohydrate units (36, 37, 55). Therefore, the ions at m/z 850, 678, 549, and 478 have sequential mass differences of 172,129, and 71 daltons which are characteristic of A2pm, Glu, and Ala, respectively. Taken together, the ions resulting from cleavage at the disaccharide end with charge retention on the peptide portion, and those from the peptide portion with charge retention on the disaccharide provide overlapping sequence information which fully characterizes the primary structure of this compound. The same interpretation scheme was employed in the assignment of the structures of five other peptidoglycan monomers which produced sufficient sequence ions to verify their structures. It should be noted, however, that none of these gave as many fragment ions as GlcNAc-(1,6-anhydro)-MurNAc-Ala-Glu-A2pm-Ala, MH' 922 (Fig. lA), even though they were all examined as single component samples at concentration levels of 5-15 nmollpl matrix.
Although the mass spectrum shown in Fig. 1A exhibits several abundant ions indicative of the structure (Fig. 2), much of the ion current is due to the chemical background (the low intensity continuum along the m/z axis) and glycerol cluster ions (labeled with an asterisk). As already mentioned, unless a large sample is used (and available), the matrix ions will often mask the sample ions of interest, and frequently only matrix-related ions can be observed below m/z 300.
In addition to the Chalaropsis and anhydro monomers from strain RD5 characterized above, Chalaropsis monomer preparations from strain FA19 known to contain a large fraction of 0-acetylated components (22,43) were analyzed by HPLC and FABMS. A total of 150 pg of peptidoglycan from strain FA19 was separated by reversed phase HPLC. A typical HPLC chromatogram for the separation of this preparation is shown in Fig. 3. The observed protonated molecular ions of the peptidoglycan monomers present are listed in Table I11 along with their primary structures which were determined by FAB MS/MS (see below). Although the mixture of Chalaropsis peptidoglycan monomers derived from strain FA19 is even more complex than that obtained from strain RD5, only five components of the former mixture had not previously been detected in the latter ( Table I)

anhydro)MurNAc-Ala-Glu-A2pm
GlcNAc-MurNAc-Ala-Glu-Aapm  Table I11 because their structures have not been proven umambiguously), have molecular weights suggesting the addition of asparagine and arginine to MH' 911 and 869, respectively (Table 111). Also observed is an ion at m/z 1036 which appears to be the same as that in strain RD5. The total yield of anhydro peptidoglycan monomers in strain FA19 Chalaropsis monomers was less than that present in the corresponding Chalaropsis monomer preparations of RD5. As in strain RD5 (Table I), glycinecontaining compounds at MH' 926 and 997 were detected.

GlcNAc-MurNAc-Ala-Glu-Azpm-Ala-Gly GlcNAc-MurNAc-Ala-Glu-A,pm-Ala-Ala
Chalaropsis monomers from strain RD5 had much fewer 0acetylated constituents than did their FA19 counterparts, consistent with previous results (22). In fact, no @acetylated derivatives were detected by FABMS in either unfractionated or HPLC-fractionated monomers from strain RD5. As in the case of peptidoglycan from strain RD5, a small fraction of Chalaropsis monomers from strain FA19 were found to contain glycine and even a smaller fraction to contain asparagine. Trace levels of some amino acids (notably glycine) in acid hydrolysates of peptidoglycan have been frequently reported but in many cases were simply attributed to contamination. It is, therefore, significant that FAB MS/MS demonstrates in an unequivocal manner that these compounds are actual constituents of gonococcal peptidoglycan fragments.
Although HPLC yielded fractions of Chaluropsis monomers from strain FA19 that contained only a single component, very little sequence information could be derived for any of the compounds containing an 0-acetyl group. It was therefore impossible to assign the position of the 0-acetyl group based on FABMS alone.
Tandem Fast Atom Bombardment Mass Spectrometry-In the normal mode of FABMS, the abundance and number of fragment ions is very dependent on several factors as discussed in the previous section. Typical of normal FAB mass spectra is the continuous background of ions associated with the matrix (Fig. 1A); this background makes it difficult to distinguish ions due to the sample from those associated with the matrix. In order to remove the contribution of the matrix, eliminate the ambiguity related to the origin of various sample ions in the mass spectrum, and increase the extent of structural information, MS/MS was employed in the analysis of peptidoglycan. The advantage of this approach is demonstrated by a comparison of the normal (Fig. 1A) and tandem (Fig. 1B) FAB mass spectra of GlcNAc-(1,6-anhydro)-

'eptidoglycan Monomers 7519
MurNAc-Ala-Glu-A2pm-Ala, MH' 922. The most important difference is the absence of the continuous background and the associated increase in the signal to noise ratio in the MS/ MS spectrum (Fig. 1B). The concentration of GlcNAc-(1,6anhydro)-MurNAc-Ala-Glu-A,pm-Ala, MH' 922, was 1 nmol/pl in Fig. lB, which is a factor of 10 less than that required to produce the spectrum shown in Fig. 1A. Furthermore, the MS/MS spectrum was acquired without prior separation of the peptidoglycan monomers by HPLC.
An important aspect of a high performance tandem mass spectrometer is the ability to select a single mass for collisioninduced dissociation. The result of this selectivity is that all the product ions observed in the MS/MS spectrum must be related to the precursor ion, MH' 922 in Fig. 1B. It should be reiterated that although the FABMS spectrum shown in Fig.  1A exhibits all the peaks that dominate Fig. 1B, one should keep in mind that peaks in the normal spectrum can also originate from either the matrix or other components of the matrix, possibilities which complicate the interpretation.
A summary of the sequence ions observed in the MS/MS spectrum and their related structural components are shown in Fig. 2 and have been discussed above in connection with

Mass Spectrometry of Peptidoglycan Monomers
A2pm, m/z 302, and A2pm, m/z 173, which provide further confirmation of the peptide sequence and indicate the location of amide groups if present. For example, in the fragment Glu-A2pm there are two possible sites of amidation, and the mass shift from m/z 302 to 303 will not define its location. However, if in conjunction with this there is a shift from m/z 173 to 172 for A2pm, then the replacement of -OH with -NH, must have occurred a t A2pm. The location of the amide group is important since it affects the somnogenecity of these compounds (15). Furthermore, in work with peptides the low mass end provides information, in the form of immonium ions, which aids in determining the amino acids present in the compound (55). In addition to these ions there are several others which are observed below mfz 300 in the MS/MS spectrum (Fig. 1B) which are not observed in the normal FABMS mass spectra (Fig. lA)  The interpretations discussed above were used to interpret the remaining FAB MS/MS spectra ( Table I).
The above data demonstrate the power of MS/MS by providing detailed sequence information for individually selected components in a mixture of structurally related peptidoglycan analogs. As previously mentioned, peptidoglycan strain FA19 contains several components whose molecular weights correspond to the addition of an 0-acetyl group, but FABMS of these HPLC-purified monomers gave only molecular weight information (thus simply confirming by homology the existence of 0-acetylated monomers). MS/MS was therefore used to determine the position of the 0-acetyl groups (Table 111). The lack of structural information from the 0acetylated compounds in the normal FAB mass spectra was due to the fact that each of these pure compounds represents only a small fraction of the total amount of starting material, and this was not sufficient to observe fragment ions. The structural assignments were based on a comparison of the MS/MS spectra of two components present in Chalaropsis monomers derived from strain FA19. They were GlcNAc-MurNAc-Ala-Glu-A2pm-Ala, MH' 940, and the compound of MH' 982 which differs from the former by 42 daltons corresponding to the replacement of -OH by -OCOCHs. A comparison of the MS/MS spectra of these compounds (Fig. 4, A and B ) indicates a number of similarities both in relative intensities and mass assignments of several ions. As discussed above, ions characteristic of loss of the disaccharide moiety are clearly observed in both compounds at m/z 534, 532, and 517. In addition to these ions, those at m/z 446,391, 302,262, and 173 are common to both spectra in Fig. 4 (and also Fig.  1B) and provide sequence ions related to the peptide portion of the molecule (Fig. 5, A and B).
Based on the observation that all the ions corresponding to the peptide portion were the same for MH+ 940 and 982 (and also for MH' 922, Fig. lB), the structural difference must be in the disaccharide portion of the molecule. An ion characteristic of GlcNAc, m/z 204, was observed in both compounds (also in Fig. I), thus eliminating it as the site of 0-acetylation. Three characteristic ions associated with the disaccharide portion of the molecule were observed a t m/z 765, 737, and 719 in Fig. 4A and are shifted by 42 daltons to m/z 807, 779, and 761, respectively, in Fig. 4B. These mass differences indicate that the 0-acetyl group was located on the MurNAc moiety, either at C-1 or at C-6. The former can be eliminated as a possibility based on the very abundant ion at MH' -18 in Fig. 4, A and B, i.e. m/z 922 and 964, respectively. This very facile loss is characteristic of a -OH group at the C-1 position of the reducing end of the disaccharide. If the molecule were acetylated at the C-1 position, a very abundant ion signal would be observed at MH' -60 ( i e . m/z 922) corresponding to the loss of the components of acetic acid in Fig.   4B, but such an ion is not observed. Based on these results, 0-acetylation has taken place at C-6 of muramic acid. This is confirmed by the peak a t m/z 689 in Fig. 4, A and B (Table 111). The presence of the 0-acetyl group on the MurNAc moiety is also supported by the observation that no 0-acetyl derivative of the 1,6-anhydro component was found in this mixture (Table 111).

DISCUSSION
These studies have employed FABMS and introduced MS/ MS as powerful tools for structural analysis of low molecular weight peptidoglycan derivatives. The objective of this comprehensive analysis of diverse peptidoglycan compounds by FABMS and MS/MS was to define chemically two complex families of biologically active disaccharide peptide monomers isolated from gonococcal peptidoglycan. As such, the results seem significant in terms of both the chemistry of peptidoglycan generally and the pathobiology of N . gonorrhoeae specifically.
From the chemical perspective, we have shown that peptidoglycans are amenable to FABMS and MS/MS and that the amount of structural information revealed depends on the instrumentation and on the quantity, purity, and uniformity of peptidoglycan. Normal FABMS can provide molecular weight information from multicomponent mixtures, allowing rapid verification of the composition of the mixture. Separation by HPLC and analysis by FABMS may provide some sequence information if tens of nanomoles of purified sample is available. If more than one component is present or if impurities such as salts are part of the mixture, little if any structural information will be obtained. Furthermore, even if large quantities of material are available, the constant matrix  Fig. 4, A and B . background will mask some of the ions associated with the sample, especially below m/z 300.
Tandem mass spectrometry is a rapid, reliable method to resolve problems concerning the structure of low molecular weight peptidoglycan derivatives, e.g. unambiguous determination of the primary sequence of amino acids and amino sugars and of the location of substituents such as 0-acetyl groups. Furthermore, in conjunction with HPLC it is possible to isolate single peptidoglycan components which retain their original, biologically active structure, to characterize these components by FABMS and MS/MS, and then to use these compounds in structure-function studies (15,30,53,54).
In the long run, the real dividends from the applications of FABMS (and especially MS/MS) to peptidoglycan chemistry will likely result from our understanding the structural basis for biological activities mediated by naturally occurring peptidoglycan derivatives, a class of compounds currently under extensive examination for their role in health and disease (56). Indeed, the recent work of Krueger et al. (53,54) defining the structural requirements for the potent somnogenic activity induced by naturally occurring peptidoglycan monomers relied on the diverse set of analogs present in preparations of gonococcal peptidoglycan monomers and the structural analysis of these compounds by FABMS. Thus, it was demonstrated (i) that the anhydromuramic acid residue (but not the glucosamine moiety) was essential for maximal somnogenic potency, (ii) that the activity was modulated by the length and composition of the peptide side chain, and (iii) that amidation of carboxyl groups on the peptide may regulate the sleep-inducing activity.
The results of the current study should also be of benefit to our studies dealing with the role of peptidoglycan in the pathobiology of gonococcal disease, specifically. Thus, it should now be possible to define the structural basis by which gonococcal peptidoglycan fragments, e.g. anhydro monomers, produce arthritis (23) and damage human fallopian tube mucosa (24). Yet, beyond somnogenic peptidoglycan compounds and gonococcal infections specifically, the enhanced capacity to define peptidoglycan structurally will, we hope, contribute to other studies concerned with the physiological role of peptidoglycan in bacteria or with biologically relevant activities resulting from peptidoglycan-host interactions.